BETA
This is a BETA experience. You may opt-out by clicking here
Edit Story

The Amazing Super Powers Of A Supercomputer

SungardAS

By Gail Dutton

Supercomputers have a Clark Kent image, but they have enough super powers underneath their nerdy exteriors to make the man of steel blush. Faster than a speeding bullet? How passé! These machines have been clocked at a whopping 33.8 petaflops. Able to leap tall buildings? Today’s supercomputers take on the biggest jobs – tasks other computer systems simply can’t handle.  It’s no wonder a supercomputer was the top choice for the National Oceanic and Atmospheric Administration (NOAA) when it upgraded its computing resources last autumn.

The reasons, explains NOAA spokesperson Ciaran Clayton, are speed and predictability. “NOAA provides the weather and climate forecasts and warnings for the protection of life and property. Our operational weather models run across thousands of processors and require computing resources that can be optimized to run predictably.” Only a supercomputer can provide that assurance.

With a Cray supercomputer, NOAA has the assurance and predictability it needs to provide the most accurate weather forecasts and warnings.

Big Science…and a Lot More

NOAA chose Cray supercomputers to meet its mission. With an annual growth rate of 15%, “Cray is going through one of the largest growth periods of any computer infrastructure provider in the industry right now,” notes Barry Bolding, VP, marketing and business development, Cray, Inc. For that, it can thank big data analytics.

As organizations need real-time analysis of increasingly complex data, they are turning to supercomputers. These machines are designed for highly-complex, real-time calculations.

“They’re used by financial services firms for market risk analyses and high-frequency trading on stock exchanges, by energy companies for gas and oil exploration, and by manufacturers to optimize manufacturing systems. One major league baseball team uses a Cray to run pre-game simulations,” Bolding says.

Super Speed

These machines are fast. To put their speed into perspective, Amazon’s EC3 cloud can process at a rate of about 1.2 petaflops. That’s 1.2 quadrillion operations per second.

Meanwhile, the world’s fastest supercomputer – NUD’s TIANHE-2, at China’s National Super Computer Center in Guangzhou – processes at a rate of 33.8 petaflops. With more than 3 million processing cores, it’s the equivalent of 750,000 modern laptops. The world’s second-fastest supercomputer, Cray’s TITAN at Oak Ridge National Laboratory, performs at nearly 17.6 petaflops, using more than 560,000 cores.

NOAA’s new computers aren’t that fast, but they’re still impressive. The recent upgrades bring its total computing capacity to 5 petaflops, a ten-fold increase from its previous capacity. That’s important because supercomputers take on projects that can’t be distributed across multiple computers.

Speed isn’t the only attraction. Another benefit, Clayton points out, is that “supercomputers are application-optimized.” Once the application is tuned to run efficiently on certain hardware, it should run reliably, the same way, every time.

And Ironman, too

Intense, steady flows of complex computations are the best uses for supercomputers. NOAA, for example, analyzes vast amounts of complex climate data, but those quantities are fairly stable. As Clayton says, “There is a predictable nature to this workload.”

That said, modern supercomputers are more scalable than the “big iron” of the 1980s and 1990s. “With Cray – but not all supercomputers – it’s easy to double or triple the size of a system and have it work as one machine. We have a very modular pathway,” Bolding says. It’s reminiscent of Ironman’s evolution from iron to a highly capable, responsive system. “The older machines were built for speed at any cost,” he continues. “Today, they’re built to optimize the architecture to match the application.”

NOAA, for example, chooses the application first and the architecture second, planning its computing advances around corresponding advancements in mission applications. The goal, Clayton says, is to “work towards providing the best platform to achieve the performance needed to accomplish the mission.”

Its mission of predicting weather accurately and to within a few miles necessitates the super powers of supercomputers. As Clayton says, “The public increasingly demands more accurate and detailed weather warnings and forecasts, which results in greater demand for supercomputing resources.”  Supercomputers, like super heroes, are popular again.

Additional Reading:

1. The Top 3 Reasons Cloud Computing Implementations Fail 

2. What Is The Real Use Of Big Data Technology In Our Society? 

3. The Best Way To Say Goodbye To Your End-Of-Life Data Center Equipment

 

POST WRITTEN BY